Meta built an AI that predicts how your brain will respond to content 🧠👀 #trendingshorts #ai

Meta's FAIR lab has released TRIBE v2, a foundation model that predicts human brain activity from video, audio, or text inputs.

Trained on over 1,000 hours of fMRI data from 720 subjects, the model's predictions of group-level brain responses outperformed most individual brain scans in accuracy.

Researchers replicated classic neuroscience experiments entirely in software, correctly identifying known regions for face recognition, language processing, and emotion.

Meta open-sourced the model weights and code. The release comes as the company expands its hardware ecosystem, including Ray-Ban smart glasses and a neural wristband, raising questions about how these tools could eventually intersect with its other ventures, particularly their advertising business.

Study: “A foundation model of vision, audition, and language for in-silico neuroscience”

Code: github.com/facebookresearch/tribev2
Weights: huggingface.co/facebook/tribev2
Demo: aidemos.atmeta.com/tribev2

Видео Meta built an AI that predicts how your brain will respond to content 🧠👀 #trendingshorts #ai канала Rowan Cheung